#589, #588, #587, #379: Optimized balanced KD-tree for C++ pyclusteri…

…ng. DBSCAN, OPTICS, CURE optimization.

ccore/include/pyclustering/cluster/dbscan.hpp

@@ -27,7 +27,7 @@
 #include <cmath>
 #include <algorithm>
 
-#include <pyclustering/container/kdtree.hpp>
+#include <pyclustering/container/kdtree_balanced.hpp>
 
 #include <pyclustering/cluster/dbscan_data.hpp>
 
@@ -60,21 +60,21 @@ Implementation based on paper @cite inproceedings::dbscan::1.
 */
 class dbscan {
 private:
-    const dataset       * m_data_ptr      = nullptr;       /* temporary pointer to input data that is used only during processing */
+    const dataset *            m_data_ptr      = nullptr;  /* temporary pointer to input data that is used only during processing */
 
-    dbscan_data         * m_result_ptr    = nullptr;       /* temporary pointer to clustering result that is used only during processing */
+    dbscan_data *              m_result_ptr    = nullptr;  /* temporary pointer to clustering result that is used only during processing */
 
-    std::vector<bool>   m_visited         = { };
+    std::vector<bool>          m_visited         = { };
 
-    std::vector<bool>   m_belong          = { };
+    std::vector<bool>          m_belong          = { };
 
-    double              m_initial_radius  = 0.0;    /* original radius that was specified by user */
+    double                     m_initial_radius  = 0.0;    /* original radius that was specified by user */
 
-    size_t              m_neighbors       = 0;
+    size_t                     m_neighbors       = 0;
 
-    dbscan_data_t       m_type            = dbscan_data_t::POINTS;
+    dbscan_data_t              m_type            = dbscan_data_t::POINTS;
 
-    container::kdtree   m_kdtree          = container::kdtree();
+    container::kdtree_balanced m_kdtree = container::kdtree_balanced();
 
 public:
     /*!

ccore/include/pyclustering/cluster/optics.hpp

@@ -28,7 +28,7 @@
 #include <set>
 #include <tuple>
 
-#include <pyclustering/container/kdtree.hpp>
+#include <pyclustering/container/kdtree_balanced.hpp>
 
 #include <pyclustering/cluster/optics_data.hpp>
 #include <pyclustering/cluster/optics_descriptor.hpp>
@@ -100,11 +100,11 @@ class optics {
 
     optics_data_t       m_type              = optics_data_t::POINTS;
 
-    container::kdtree   m_kdtree            = container::kdtree();
+    container::kdtree_balanced      m_kdtree            = container::kdtree_balanced();
 
-    optics_object_sequence *            m_optics_objects    = nullptr;
+    optics_object_sequence *        m_optics_objects    = nullptr;
 
-    std::list<optics_descriptor *>      m_ordered_database  = { };
+    std::list<optics_descriptor *>  m_ordered_database  = { };
 
 public:
     /*!

ccore/include/pyclustering/container/kdnode.hpp

@@ -1,51 +1,53 @@
-/**
-*
-* @authors Andrei Novikov ([email protected])
-* @date 2014-2020
-* @copyright GNU Public License
-*
-* GNU_PUBLIC_LICENSE
-*   pyclustering is free software: you can redistribute it and/or modify
-*   it under the terms of the GNU General Public License as published by
-*   the Free Software Foundation, either version 3 of the License, or
-*   (at your option) any later version.
-*
-*   pyclustering is distributed in the hope that it will be useful,
-*   but WITHOUT ANY WARRANTY; without even the implied warranty of
-*   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-*   GNU General Public License for more details.
-*
-*   You should have received a copy of the GNU General Public License
-*   along with this program.  If not, see <http://www.gnu.org/licenses/>.
-*
+/*!
+
+@authors Andrei Novikov ([email protected])
+@date 2014-2020
+@copyright GNU Public License
+
+@cond GNU_PUBLIC_LICENSE
+    pyclustering is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    pyclustering is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+@endcond
+
 */
 
 
 #pragma once
 
 
+#include <functional>
 #include <memory>
 #include <vector>
 
+#include <pyclustering/definitions.hpp>
+
 
 namespace pyclustering {
 
 namespace container {
 
-/**
- *
- * @brief   Node of KD Tree.
- *
- */
-class kdnode {
+/*!
 
-friend class kdtree;
+@brief   Node of KD Tree.
 
+*/
+class kdnode : public std::enable_shared_from_this<kdnode> {
 public:
-  using ptr         = std::shared_ptr<kdnode>;
+  using ptr                 = std::shared_ptr<kdnode>;
 
 private:
-  using weak_ptr    = std::weak_ptr<kdnode>;
+  using weak_ptr            = std::weak_ptr<kdnode>;
+  using search_node_rule    = std::function< bool(const kdnode&) >;
 
 private:
     std::vector<double>           m_data    = { };
@@ -68,13 +70,17 @@ friend class kdtree;
 
     virtual ~kdnode() = default;
 
-private:
+public:
     void set_left(const kdnode::ptr & p_node);
 
     void set_right(const kdnode::ptr & p_node);
 
     void set_parent(const kdnode::ptr & p_node);
 
+    void set_data(const point & p_data);
+
+    void set_payload(void * p_payload);
+
     void set_discriminator(const std::size_t disc);
 
 public:
@@ -84,7 +90,32 @@ friend class kdtree;
 
     kdnode::ptr get_parent() const;
 
-    void * get_payload();
+    void * get_payload() const;
+
+    /*!
+
+    @brief   Find node in KD-tree using current node as a root of a subtree, the first founded node with
+    specified coordinates is returned.
+
+    @param[in] p_point: coordinates of searched node.
+
+    @return  Pointer to found node in the sub-tree.
+
+    */
+    kdnode::ptr find_node(const point & p_point);
+
+    /*!
+
+    @brief   Find node using specified rule, it returns the first node that satisfy search rule.
+
+    @param[in] p_point: coordinates of searched node.
+    @param[in] p_cur_node: node from which search is performed.
+    @param[in] p_rule: rule that should be satisfied by searched node.
+
+    @return  Return the smallest node in specified subtree in line with discriminator.
+
+    */
+    kdnode::ptr find_node(const point & p_point, const search_node_rule & p_rule);
 
     const std::vector<double> & get_data() const;